I was just using that as an example, though I'm not sure 14 million would actually be too few to have that kind of separation of responsibilities. Assuming no genocide or catastrophe, by the time human population gradually dwindled back down to that level, we'd have hopefully settled on a fairly sustainable and automated existence.

The rise of AI seems to suggest otherwise.

This is still a very shaky ticket to bet on. Yes, the new crop of AI tools promises a lot of productivity gains, but for now they also need well-educated folks to operate them and integrate their output, and we don't quite know what the reliance of AI will do to peoples' educational interest/drive, etc. There might well be an overall downward spiral and loss of operating knowledge lurking there.

It would mean we reached peak capability as a civilization, and then we would be forced to rely on an artifact that can pull from that subset of knowledge to help patch together machines that broke down until we lost all the hardware to run the AI itself.